Comparison of Time-Varying Load Models for Estimating CVR Factor and VSF Using Dual-Stage Adaptive Filter

Abstract

Assessment of conservation voltage reduction (CVR) deployment largely depends on load-to-voltage sensitivity (LTV). Therefore, two quantifiers—voltage and the load consumption—are truly responsible for measuring the CVR factor (CVRF), which is the primary index for calculating the load or energy consumption reduction. Simultaneously, CVR deployment improves the voltage profile by providing line relief due to the reduction in consumption. Hence, it is also important for the utilities to know the voltage sensitivity factor (VSF) during the CVR deployment period. This paper provides a comprehensive comparison of two different voltage sensitive load models—ZIP and exponential model—to estimate the CVRF and the VSF. The comparison is done in a time varying manner since the parameters of load models vary with time as the voltage changes. Two different sets of dual-stage filtering methods are used to estimate the load parameters and observe which load model accurately identifies the CVRF and the VSF. Stochasticity in load variation is also considered in the estimation process. This process helps the utilities to assess the CVRF and VSF for a target feeder using the most accurate load model. The entire estimation is done using actual data from a 22.86 kV distribution feeder, provided by duke energy progress. Moreover, the validation of the estimated parameters is done using the original circuit associated with the measurements in OpenDSS Software.